Last week, I had the pleasure of joining the finals of a startup competition organized by Oleg Mosyazh in collaboration with The Ventures by Daria Kuk. The event was everything a founder or investor could hope for: strong startup lineup, solid traction, beautiful venue, and a world-class jury.
But I wasn’t there just to watch or judge.
I came with a different goal: to test our AI-powered VC Analyst tool — PitchBob — in a real-world, high-stakes environment. Specifically, I wanted to see how our tool’s pitch deck evaluations compared to the actual decisions made by experienced venture investors.
Spoiler: the results were eye-opening.
Why We Did It
Evaluating startups is hard — especially at scale. Every week, accelerators, VCs, angels, and corporates receive hundreds of decks. Most decisions are based on a quick scan or gut feel. That leaves a lot of room for bias, inconsistency, and missed opportunities.
Our mission at PitchBob is to democratize structured, high-quality feedback, using AI to simulate a venture analyst’s thought process. This event gave us a unique opportunity to:
- Benchmark our AI scores against human investor decisions
- Explore how stage performance aligns (or not) with submitted decks
- Understand how founders present themselves differently in writing vs. live
How the Experiment Worked
We partnered with the organizers to analyze every pitch deck submitted to the competition — both from early applicants and the 10 finalists. Using our AI-powered tool, each deck was evaluated on 11 core VC-style criteria, such as:
- Problem/Solution fit
- Market size and growth
- Team strength
- Business model
- Competitive advantage
- Go-to-market strategy
- Traction and KPIs
- Exit potential
Each deck was scored on a 0–100 scale, and we generated:
- A structured investment memo
- Executive summary
- Pros and cons of the opportunity
- Red flag alerts
- Personalized feedback messages
The goal wasn’t to pick winners — it was to observe patterns and see how our AI analysis compared to the actual decisions made by the jury after live pitches.
PitchBob’s Top 3 Startups
Based on our AI scoring, the top decks were:
- IMMA Health – 81
- Adversa AI – 80
- Senecio – 79
These teams stood out in terms of:
- Clear, well-articulated problem and solution
- Logical structure and visual clarity
- Strong teams with domain expertise
- Credible go-to-market approaches and early traction
We intentionally didn’t influence the jury or discuss the AI results beforehand.
And the Jury Said…
After watching the pitches live, here’s how the jury awarded the startups:
- Best Product – Senecio
- Best Traction – Puls Project
- Best Team – IMMA Health
- Best AI Model – AI Light
There was strong alignment between the AI results and human judgment in the top categories — especially for Senecio and IMMA Health, which ranked highly in both.
One surprise: Adversa AI, our #2, didn’t receive a jury award — possibly due to differences in how the team presented live, or because of more subjective preferences among the judges.
Where Human and AI Evaluation Diverged
There were also cases where AI and jury results differed. For instance, AI Light, which won “Best AI Model” from the jury, received a lower score from our tool. Why?
From our AI’s perspective:
- The deck lacked clarity on the go-to-market plan
- The business model was underdeveloped
- Several key metrics were missing or vague
This led to a modest overall score — even though the technology itself may have wowed the jury live.
This gap highlights one of our key hypotheses:
Founders often “oversell” on stage and “undersell” in their decks.
In other words, the email-submitted deck may be a more honest version of the startup — one that lacks charisma but shows the real level of maturity.
What We Learned
1. Correlation Exists
There’s real overlap between AI-generated scores and experienced investor judgments.
This validates the model’s underlying logic and its ability to surface high-potential startups with no human bias.
2. Pitch ≠ Deck
What’s on stage often differs from what’s in the slides. Decks are a static document — but they represent the founder’s clarity and planning skills. Live pitches, by contrast, reflect energy, storytelling, and confidence.
3. Real-time Analysis = Real-time Insight
Using PitchBob live during the event made it easier to validate assumptions, generate follow-up questions, and quickly identify strengths and red flags.
We’re now working on a “Smart Follow-up Prompt Generator” that will suggest intelligent questions based on what’s missing or weak in a deck.
Founder Reactions
Several startup founders approached us after the event to ask for their PitchBob scores and feedback. One said:
“This is the kind of structured feedback we never get from VCs. Usually, it’s just silence — or one vague sentence.”
Another added:
“I didn’t realize how unclear our GTM strategy was until I saw the AI’s breakdown. We’ll definitely fix that before our next round.”
That’s exactly the kind of impact we’re building for.
What’s Next: New Features and Use Cases
Based on this live test, we’re already working on several upgrades:
- Live Pitch vs. Deck Delta: A feature that compares the emailed deck with the on-stage pitch (or transcript) to assess message consistency.
- “Confidence Score”: To highlight how strong and coherent a founder’s narrative is across channels.
- Smart Follow-Up Generator: Suggesting key investor-style questions to ask after reviewing a deck.
- Real-Time Demo Day Dashboards: So accelerators and VCs can monitor startup scores in real time during demo days.
- Public Scoreboards and Voting: For community-driven feedback at pitch competitions.
We’re also expanding our partnerships with accelerators, universities, and corporate venture arms who want a second pair of eyes — or hundreds of them — for pitch evaluation and training.
Try It Yourself
If you’re reviewing dozens of pitch decks a month and want a faster, more structured way to prioritize and give feedback — PitchBob is here to help.
We’re still in open beta, offering free evaluations for up to 10 decks.
Test it here.
Final Thought
AI is not here to replace investors — it’s here to augment them.
What we saw at this event is just the beginning: tools like PitchBob can bring speed, clarity, and structure to a process that is often chaotic, inconsistent, and biased.
If you’re building the future, evaluating the future, or just trying to keep up with your inbox full of decks — let’s talk.
